Self-reported adverse events in health care that cause harm: a population-based survey

Adverse events in health care are common.1,2 Most current knowledge of adverse events is based on reviews of hospital medical records,1,2 incident reports by health staff3,4 or analysis of administrative databases.5 These approaches each have strengths, but also inherent biases and weaknesses,5,6 as many events will go unreported and unrecorded.7 Comparatively little is known about adverse events outside hospitals, although some evidence suggests they may be a significant contributor to harm in health care.8

Health care consumers are a relatively underused source of information about adverse events9 and about their views about such events. This underuse occurs despite evidence of consumers’ capability in noticing adverse events.9,10 Patient satisfaction surveys tend not to focus on adverse events so much as problems with interpersonal interactions or the delivery of care, and are most commonly conducted among inpatients.

An Australian population survey in 2000 reported an annual incidence of adverse events causing harm of 6.5% among adults in the community;11 these results are similar to those of a United States study.12 However, the Australian study’s low response rate of 43% raises concerns about response bias. It is unclear whether the rate of occurrence of adverse events has changed since the 2000 survey,11 in the wake of well publicised scandals13 and with the increased focus among clinicians, health care providers and policymakers on patient safety.13

We examined the incidence of self-reported adverse events causing harm in health care within a representative population sample of community-dwelling adults. We also used survey items from the previous Australian survey11 to examine attitudes about the performance of the health system in relation to safety.

The North West Adelaide Health Study (NWAHS) is a biomedical population study of a representative sample of people aged 18 years or older, randomly selected from the electronic White Pages telephone directory and living in the north-western suburbs of Adelaide, South Australia (regional population, 0.6 million).

In 2000, 4060 adults (1988 men and 2072 women; 69.4% of those who completed the initial interview) attended the clinic for biomedical examination. In 2004 and 2005, all participants were contacted again to be interviewed in Stage 2 of the NWAHS; 3522 (86.7%) completed these interviews, which provide the data for this study. The methods of the NWAHS,14 and the validity of these methods of selection in achieving an unbiased sample have been described previously.15

This study provides a recent picture of what a representative Australian population sample reports about the safety of their health care. The annual incidence of harmful adverse events as perceived by participants in our study was significantly lower than those in earlier Australian and US studies that used identical questions.11,12 Harmful adverse events have consequences that linger, with nearly a third of the respondents in our study who reported adverse events causing harm reporting long-term or permanent emotional effects, and over 20% describing long-term physical consequences.

The likelihood of experiencing harm from an adverse event over the past year was independently related to recent hospitalisation in that year, suggesting an increased risk associated with more complex disease and treatment. These findings further justify the focus of efforts to reduce adverse events in hospitals, and after discharge from hospital, as appropriate.18 Some individual characteristics (but not age) also influenced the association with reporting adverse events. Participants with either low income or higher education reported more adverse events. Higher education may increase the chances of individuals recognising problems with their care, or alternatively may be associated with increased expectations of care processes and outcomes.

The other independent factor associated with adverse events was a tendency to risk-taking behaviour. Such behaviour is associated with an increased chance of injury in different contexts.19 Willingness to take risks differs with context. However, the correlation across contexts is strong, suggesting that a single trait of risk-taking attitude exists.20,21 There is evidence to indicate that differences that exist across situational contexts are more likely to be differences in risk perception than in risk preferences.21 The significance of this for practitioners and the health care system lies in the potential for altering risk perception and risk-taking behaviour by improving communication. Communicating in a way that focuses on improving patients’ ability to perceive risk more accurately might reduce the incidence of adverse events. This may be as simple as clinicians and patients actively discussing and acting on such things as medication-related symptoms.22

The frequency of harmful adverse events in our study is not comparable with those in studies identified by medical record review or other methods of administrative data collection. Such studies sample records of patient admissions, whereas our study included non-hospital settings.

The reliance on patient reports of adverse events is a potential limitation of our study, as these reports were not externally validated or subject to any objective examination. Self-report can create problems with attribution, as respondents may confuse illness-related experiences with health care-related adverse events. It is also unclear how the health beliefs of individuals may influence their perceptions of the health system and how experiences will be reported.

Recall bias in the reporting of adverse events may also have been a limitation of our study. The level of satisfaction with recent health care was significantly related to the reporting of adverse events. We do not have data to infer the direction of causality — that is, does an adverse event reduce satisfaction with care, or are those less satisfied more likely to notice or perceive experiences as adverse events? However, previous studies have shown patient reports of problems are similar to those found from other data collection methods.23 A combination of underreporting by some patients and overreporting by others may account for this concordance.

Substantial variation can exist between event rates estimated from different sources of external data, such as medical records, staff incident reports or administrative databases.24 Patient reports thus become an alternative view on system quality, and allow us to identify potential target areas for intervention. Self-reports provide opportunities to explore patients’ perceptions of their care and the health system in collaborative reflective dialogue.

Our study was also subject to potential bias from survey non-response, although response rates in our sample were relatively high.25 In particular, there were no major differences between NWAHS participants and the general population in terms of health indicators or lifestyle behaviours.26 The strength of our study is the large representative population sample from which we obtained self-reported adverse events, covering both inpatient and ambulatory care.

A recent report indicated that Australians trusted their general practitioner more than other people involved in the health care system.27 The issue of the importance of trust placed in an individual has a corollary in the attitudes expressed about which groups affect patient safety. People were more likely to consider individual health care professionals as having a positive effect on safety than professional groups or those overseeing the health care system (eg, governments). Also, most respondents indicated that more resources were the key to improving the safety of health care. The converse of this can be seen in the views expressed about problem issues in health, which were focused around negligence and getting the right treatment. Again, this reflects a view that responsibility lies mostly with the individual health care professional rather than in systems improvement.

Despite many articles in the professional literature,28 there has been relatively little in the way of public health campaigns that put the view that systems change is the key to patient safety. Public discourse has often focused on the “bad apples” notion of safety, as evidenced by the publicity surrounding a small number of notorious cases.13,28 Some jurisdictions are advocating incident reporting and public reporting of health care outcomes in hospitals as a means of shifting the focus away from individual blame and onto the action being taken to improve hospitals’ performance.29 However, despite widespread acceptance in health care circles,5,13,28 a general public and political consensus has not been reached that supports a systems approach to patient safety. We feel it is time that professional bodies and public health care workers take up the task of informing and engaging with consumers to build a consensus to exert political pressure, so that systems improvement will eventually lead safety improvement.

1 Perceived physical, emotional and financial effects among the 148 participants who reported adverse events causing harm